Analysis Of User Needs For Online Reading In The Second Grade

The "China Digital Reading Market Research Report 2019" mentioned that the number of digital reading users in China in 2019 has reached 740 million,a year-on-year increase of 1.4%.Facing the complicated network world and the mixed books,it is very important for users to really read the books they need.Exploring user needs is one of the important ways to help users understand their inherent needs.Internet reading users come from all corners of the world,with different economic and cultural backgrounds,and the needs for different types of reading vary from person to person.This article uses the online reading comprehension test developed by an Internet education company as a material for the second grade primary school to investigate the second grade primary school subjects in 19 provinces,municipalities and autonomous regions across the country,and retrieved 1,309 valid data.The first study of this article aims to classify the sample population scientifically and reasonably through statistical methods.Using k-mediods clustering,latent category analysis,Knearest neighbor algorithm,discriminant analysis,and logistic regression regression analysis to explore the degree of matching with the purpose and data form of this study,mainly using R language and Mplus to analyze the data to obtain the most suitable Data analysis method for objective classification of data sets.Study 2 uses the classification results of Study 1 to test the functional differences of socioeconomic status based on the reading and answering behaviors of second-year online reading users.At the same time,the traditional economic classification indicators are used to divide the population into DIF tests to explore the reading ability The differences are compared and the results of the two grouping methods are compared.Finally,the main socio-economic status factors that affect reading performance are explored,and targeted solutions are proposed.The results of Study 1 show that: k-mediods clustering and latent category analysis are the most widely used methods in data science for population or other classifications,with minimal assumptions and restrictions on the data set.It is also very applicable to the sample data of this study,and the classification results obtained by the two methods are the same,which can better divide the population into two categories of high economic level and low economic level.The results of Study 2 show that: The population obtained by the statistical method is more authentic than the traditional classification indicators,and the DIF test results can also show regularity.Second-year online reading users mainly show reading in terms of language foundation and understanding inference.Differences in socioeconomic levels of families,family education investment and language environment in family SES become the most important influencing factors.Conclusion: When discussing the differences in reading ability of different populations,using clustering analysis,latent category analysis,K-nearest neighbor algorithm,discriminant analysis,logistic regression regression and other machine learning algorithms to classify can get more consistent results,and then apply DIF test Discussing the response of category groups can effectively identify group differences.